Indrayudh Ghosal

About Indrayudh Ghosal

Applied Scientist with 5+ years of experience in relevance ranking and recommendation systems at Amazon Search, and a PhD in Statistics from Cornell University specialising in interpretable machine learning. Build and deploy ML models in production trained on terabyte-scale data, design A/B testing frameworks that increase experimentation velocity, and drive cross-team technical standards. Research published in JCGS and Statistical Science (90+ citations).

Experience

1. Applied Scientist Ii

   Amazon

   Aug 2021 — Present · Seattle, WA, US

   Train and deploy production relevance ranking models integrating deep-learned features (withTensorFlow) into low-latency GBDT/XGBoost layers that feed downstream re-rankers. Designand evaluate personalisation features and metrics to improve product discovery. Experimentscontributed to an estimated $2B+ in incremental revenue.Developed interleaving experiments with 62x sensitivity improvement, enabling experimentationin low-data international markets with fewer customers and shorter runtimes.Consolidate redundant models worldwide, reducing global model count by 60%+ and optimisingmodel size, saving an estimated $1M+ in infrastructure costs, eliminating months of redundantscientist effort, and reducing tech debt and stale-model risk in production.Own end-to-end experimentation infrastructure (training, offline eval, A/B testing, deploymentvia CI/CD) on terabyte-scale data using Python, PySpark, SageMaker, EMR. Serve as cross-teamtechnical resource: define training standards, review production model quality, and advise scientistsacross levels on search ranking architecture.

Education

• Indian Statistical Institute, Kolkata

  Bachelor of Statistics, Mathematical Statistics, Computer Science and Probability

• Indian Statistical Institute, Kolkata

  Master of Statistics, Mathematical Statistics and Probability

• Cornell University

  Doctor of Philosophy (PhD), Statistics and Data Science

Skills

• Research
• C++
• Statistics
• Microsoft Excel
• Latex
• R
• Statistical Modeling
• C
• Customer Service
• Data Analysis
• Matlab
• Microsoft Office
• Analytics